Most phenotypic variations, including those involved in complex diseases such as ankylosing spondylitis[unreadable] (AS) and differences in drug response, are generated by integrated actions of multiple genetic and[unreadable] environmental factors. Existing methods may provide tools for analysis, but with the imminent completion of[unreadable] the HapMap Project providing a comprehensive catalogue of millions of SNPs and haplotypes across diverse[unreadable] populations and rapid development of high throughput genotyping technologies, a paradigm shift for genetic[unreadable] studies of complex traits from individual marker analysis to genome-wide association studies and systemslevel[unreadable] analysis is indispensable. Genome-wide association studies and systems-level analysis for complex[unreadable] diseases raise great challenges in three aspects. First, it is practically impossible to ensure a genome-wide[unreadable] significance level of 0.05 for testing millions of SNPs using traditional statistic methods. Second, most[unreadable] phenotypic variations are generated by integrated actions of multiple genetic and environmental factors[unreadable] through complex interactions between genes, and between gene and environments. Detecting interactions[unreadable] among genes or SNP markers is a daunting task. Third, most existing analytic methods analyze each marker[unreadable] (or haplotype) and phenotype individually, and do not consider network structures among multiple[unreadable] phenotypes and multiple markers. Therefore, new techniques need to be proposed to address these[unreadable] challenging tasks. The overall goal of this project is (1) to develop nonlinear statistics for genome-wide[unreadable] association studies for ensuring genome-wide significance levels, (2) to develop novel statistical methods for[unreadable] detection of gene interaction and efficient computational algorithms for construction of genetic interaction[unreadable] networks, (3) to develop a conceptual framework for network modeling of multiple phenotypes, and (4) to[unreadable] develop or adopt novel statistical methods for joint analysis of multiple phenotypes and multiple markers. AS[unreadable] is a complex disease. Genotype and phenotype data from projects 1-3 for dissecting complex genetic[unreadable] architecture of AS will be used for development and evaluation of methodology, and real data analysis.